jina multimodal embedding #922

Closed
opened 2026-02-16 10:21:01 -05:00 by yindo · 1 comment
Owner

Originally created by @nicho2 on GitHub (Jan 7, 2026).

Self Checks

  • This is only for bug report, if you would like to ask a question, please head to Discussions.
  • I have searched for existing issues Dify issues & Dify Official Plugins, including closed ones.
  • I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
  • [FOR CHINESE USERS] 请务必使用英文提交 Issue,否则会被关闭。谢谢!:)
  • Please do not modify this template :) and fill in all the required fields.

Dify version

1.11.1

Plugin version

0.0.14

Cloud or Self Hosted

Self Hosted (Docker)

Steps to reproduce

I've updated the jina plugin and now the embedding don't work
I used the jina-embeddings-v4 embedding (from pipeline)

I don't now if it's the solution but this model don't appear in _position.yaml file -->

  • jina-clip-v2
  • jina-clip-v1
  • jina-embeddings-v3
  • jina-embeddings-v2-base-de
  • jina-embeddings-v2-base-zh
  • jina-embeddings-v2-base-en

My connection to jina is well

Image

✔️ Error log

No response

Originally created by @nicho2 on GitHub (Jan 7, 2026). ### Self Checks - [x] This is only for bug report, if you would like to ask a question, please head to [Discussions](https://github.com/langgenius/dify/discussions/categories/general). - [x] I have searched for existing issues [Dify issues](https://github.com/langgenius/dify/issues) & [Dify Official Plugins](https://github.com/langgenius/dify-official-plugins/issues), including closed ones. - [x] I confirm that I am using English to submit this report (我已阅读并同意 [Language Policy](https://github.com/langgenius/dify/issues/1542)). - [x] [FOR CHINESE USERS] 请务必使用英文提交 Issue,否则会被关闭。谢谢!:) - [x] Please do not modify this template :) and fill in all the required fields. ### Dify version 1.11.1 ### Plugin version 0.0.14 ### Cloud or Self Hosted Self Hosted (Docker) ### Steps to reproduce I've updated the jina plugin and now the embedding don't work I used the jina-embeddings-v4 embedding (from pipeline) I don't now if it's the solution but this model don't appear in _position.yaml file --> - jina-clip-v2 - jina-clip-v1 - jina-embeddings-v3 - jina-embeddings-v2-base-de - jina-embeddings-v2-base-zh - jina-embeddings-v2-base-en My connection to jina is well <img width="1027" height="106" alt="Image" src="https://github.com/user-attachments/assets/8b24440a-5c07-41ee-9254-420ddac729f9" /> ### ✔️ Error log _No response_
yindo added the bug label 2026-02-16 10:21:01 -05:00
yindo closed this issue 2026-02-16 10:21:01 -05:00
Author
Owner

@dosubot[bot] commented on GitHub (Jan 7, 2026):

The reason 'jina-embeddings-v4' isn't working is that, while the plugin code and model config files for this model exist and support was added in plugin versions after 0.0.7, the model must also be listed in the _position.yaml file to be available for use in Dify. If it's missing from that file, Dify won't recognize or allow you to select it—even if the code supports it and your connection to Jina is fine [source 1] [source 2].

To resolve this, add jina-embeddings-v4 to the text_embedding section of your plugin's _position.yaml file, matching the format of the other models. After editing, restart your plugin and Dify services. This should make the model selectable and usable. If you still have issues, double-check that your plugin version is at least 0.0.8 (when v4 support was added), and that your pipeline or knowledge base type supports this model—pipeline-based KBs have stricter validation and may reject some models [source].

If you want multimodal (image+text) embedding, make sure the model's schema advertises the VISION feature; otherwise, Dify won't enable image input for recall, even if the model is technically multimodal [source].

To reply, just mention @dosu.


How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other

Open in Cursor Join Discord Share on X

@dosubot[bot] commented on GitHub (Jan 7, 2026): <!-- Answer --> The reason 'jina-embeddings-v4' isn't working is that, while the plugin code and model config files for this model exist and support was added in plugin versions after 0.0.7, the model must also be listed in the `_position.yaml` file to be available for use in Dify. If it's missing from that file, Dify won't recognize or allow you to select it—even if the code supports it and your connection to Jina is fine [[source 1]](https://github.com/langgenius/dify-official-plugins/pull/1263) [[source 2]](https://github.com/langgenius/dify-official-plugins/issues/2361). To resolve this, add `jina-embeddings-v4` to the `text_embedding` section of your plugin's `_position.yaml` file, matching the format of the other models. After editing, restart your plugin and Dify services. This should make the model selectable and usable. If you still have issues, double-check that your plugin version is at least 0.0.8 (when v4 support was added), and that your pipeline or knowledge base type supports this model—pipeline-based KBs have stricter validation and may reject some models [[source]](https://github.com/langgenius/dify/issues/27294). If you want multimodal (image+text) embedding, make sure the model's schema advertises the `VISION` feature; otherwise, Dify won't enable image input for recall, even if the model is technically multimodal [[source]](https://github.com/langgenius/dify/issues/30401). <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot).* --- <sup>How did I do? [Good](https://app.dosu.dev/response-feedback/88329c95-c9fe-4d86-8fd3-8fa16f250536?feedback_type=great_response) | [Irrelevant](https://app.dosu.dev/response-feedback/88329c95-c9fe-4d86-8fd3-8fa16f250536?feedback_type=irrelevant_answer) | [Incorrect](https://app.dosu.dev/response-feedback/88329c95-c9fe-4d86-8fd3-8fa16f250536?feedback_type=incorrect_sources) | [Verbose](https://app.dosu.dev/response-feedback/88329c95-c9fe-4d86-8fd3-8fa16f250536?feedback_type=too_verbose) | [Hallucination](https://app.dosu.dev/response-feedback/88329c95-c9fe-4d86-8fd3-8fa16f250536?feedback_type=hallucination) | [Report 🐛](https://app.dosu.dev/response-feedback/88329c95-c9fe-4d86-8fd3-8fa16f250536?feedback_type=bug_report) | [Other](https://app.dosu.dev/response-feedback/88329c95-c9fe-4d86-8fd3-8fa16f250536?feedback_type=other)</sup> [![Open in Cursor](https://dosu.dev/dosu-in-cursor.svg)](https://cursor.com/link/prompt?text=The%20reason%20%27jina-embeddings-v4%27%20isn%27t%20working%20is%20that%2C%20while%20the%20plugin%20code%20and%20model%20config%20files%20for%20this%20model%20exist%20and%20support%20was%20added%20in%20plugin%20versions%20after%200.0.7%2C%20the%20model%20must%20also%20be%20listed%20in%20the%20%60_position.yaml%60%20file%20to%20be%20available%20for%20use%20in%20Dify.%20If%20it%27s%20missing%20from%20that%20file%2C%20Dify%20won%27t%20recognize%20or%20allow%20you%20to%20select%20it%E2%80%94even%20if%20the%20code%20supports%20it%20and%20your%20connection%20to%20Jina%20is%20fine%20%5B%5Bsource%201%5D%5D%28https%3A//github.com/langgenius/dify-official-plugins/pull/1263%29%20%5B%5Bsource%202%5D%5D%28https%3A//github.com/langgenius/dify-official-plugins/issues/2361%29.%0A%0ATo%20resolve%20this%2C%20add%20%60jina-embeddings-v4%60%20to%20the%20%60text_embedding%60%20section%20of%20your%20plugin%27s%20%60_position.yaml%60%20file%2C%20matching%20the%20format%20of%20the%20other%20models.%20After%20editing%2C%20restart%20your%20plugin%20and%20Dify%20services.%20This%20should%20make%20the%20model%20selectable%20and%20usable.%20If%20you%20still%20have%20issues%2C%20double-check%20that%20your%20plugin%20version%20is%20at%20least%200.0.8%20%28when%20v4%20support%20was%20added%29%2C%20and%20that%20your%20pipeline%20or%20knowledge%20base%20type%20supports%20this%20model%E2%80%94pipeline-based%20KBs%20have%20stricter%20validation%20and%20may%20reject%20some%20models%20%5B%5Bsource%5D%5D%28https%3A//github.com/langgenius/dify/issues/27294%29.%0A%0AIf%20you%20want%20multimodal%20%28image%2Btext%29%20embedding%2C%20make%20sure%20the%20model%27s%20schema%20advertises%20the%20%60VISION%60%20feature%3B%20otherwise%2C%20Dify%20won%27t%20enable%20image%20input%20for%20recall%2C%20even%20if%20the%20model%20is%20technically%20multimodal%20%5B%5Bsource%5D%5D%28https%3A//github.com/langgenius/dify/issues/30401%29.)&nbsp;[![Join Discord](https://img.shields.io/badge/join-5865F2?logo=discord&logoColor=white&label=)](https://go.dosu.dev/discord-bot)&nbsp;[![Share on X](https://img.shields.io/badge/X-share-black)](https://twitter.com/intent/tweet?text=%40dosu_ai%20helped%20me%20solve%20this%20issue!&url=https%3A//github.com/langgenius/dify-official-plugins/issues/2376)
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langgenius/dify-official-plugins#922